local-openid: Single User, Ephemeral OpenID Provider

local-openid allows users with shell accounts on servers to authenticate with OpenID consumers by editing a YAML file in their home directory instead of authenticating through HTTP/HTTPS.

  1. Encounter a login page that accepts OpenID (the consumer)

  2. Login into your own server (if you're not already logged in)

  3. Start the local-openid app on your server

  4. Login using your OpenID (on the consumer)

  5. edit ~/.local-openid/config.yml on your server to approve the consumer This config file is only created after the consumer makes a successful request to your server.

  6. Reload the local-openid page your browser was on.

  7. Shut down the local-openid application.

local-openid exists for the following reasons:

  1. Passwords and password managers feel clumsy to me on web browsers. On the other hand; using ssh, editing text files, and running servers are second nature. Clearly, local-openid is not for everyone.

  2. Identity providers may not last. Companies die and business plans change. I'd rather my online identity not be subject to those whims.

  3. OpenID providers could be compromised without disclosure. With local-openid, I have server logs to know if somebody is even trying something fishy with my identity. The vector for compromising my identity is greatly reduced because my local-openid instance has 99.999% downtime.


The following command should install local-openid and all dependencies:

gem install local-openid

setup.rb is also provided for non-Rubygems users.


local-openid is a small Sinatra application. It requires the Ruby OpenID library (2.x), Sinatra (1.0), Rack and any Rack-enabled server. To be useful, it also depends on having a user account on a machine with a publically-accessible IP and DNS name to use as your OpenID identity.


"local-openid" should be installed in your $PATH by RubyGems or setup.rb. It is a Sinatra application and takes the usual command-line arguments. It binds on all addresses ( and port 4567 by default, using the standard WEBrick web server.

You may specify a different port with the -p switch and address with the -o switch. The following command will start local-openid on port 3000 bound to localhost (useful if behind a reverse proxy like nginx).

local-openid -o -p 3000


I don't have any plans for more development with local-openid. It was after all, just a weekend hack. It does what I want it to and nothing more.

You can use the mailing list to share ideas, patches, pull requests with other users. Remember, I wrote local-openid because I find the web difficult to use. So I'll only accept communication about local-openid via email :)

Feel free to fork it and customize it to your needs. Of course, drop me a line if you fix any bugs or notice any security holes in it.

You can get the latest source via git from the following locations:

git:// (mirror) (mirror)

You may browse the code from the web and download the latest tarballs here:


There is NO WARRANTY whatsoever, implied or otherwise. OpenID may not be the best choice for dealing with security-sensitive data, and this application is just a weekend hack with no real security auditing. On the other hand, it's quite hard for somebody to steal your OpenID credentials when your provider implementation has 99.999% downtime :)


mail archives:
source code: git clone
	torsocks git clone http://ou63pmih66umazou.onion/local-openid.git